Abstract Deep brain stimulation (DBS) is one of the most effective treatments for patients with advanced Parkinson's disease (PD). Delivery of high frequency electrical stimulation to the subthalamic nucleus (STN) ameliorates parkinsonian motor symptoms, often within seconds, but therapeutic effects wear off quickly if stimulation is stopped, often within minutes. This transient nature of symptomatic relief underscores the fact that existing DBS protocols mask symptoms but do not alleviate underlying circuit dysfunction. A modified DBS protocol, called coordinated reset (CR-DBS), has shown potential to provide long-lasting therapeutic benefits for days to weeks after stimulation, but this protocol has been slow to translate into widespread clinical use because (1) the multi- site, pseudorandom stimulation patterns required to implement it cannot be delivered with existing devices and (2) its mechanisms of action remain obscure, hindering insights into what parameters of CR-DBS should be tuned to ensure engagement of long-lasting effects. Recently, in a mouse model of PD, we discovered a cellular- based strategy to induce long-lasting motor recovery, by using optogenetics to target interventions to specific neuronal subpopulations in the external globus pallidus (GPe), an anatomical neighbor of the STN. Long-lasting motor rescue was induced by interventions that simultaneously increased the firing rates of GPe neurons enriched in parvalbumin (PV-GPe) and decreased the firing rates of GPe neurons enriched in lim homeobox 6 (Lhx6-GPe). Interestingly, at the physiological level, these cell-type specific interventions in the GPe converged upon a similar mechanism as CR-DBS, by ameliorating pathological patterns of neural activity in basal ganglia output nuclei that have been associated with parkinsonian motor deficits. This proposal will use knowledge gained from our discovery of long-lasting rescue through cell-type directed interventions in GPe to guide rational design and interrogation of human-applicable forms of DBS that may yield similarly long-lasting therapeutic benefit. Our experiments will test a novel, mechanistic hypothesis, based on supporting preliminary data, that the pattern of electrical DBS can be tuned to drive cell-type specific responses in the GPe that mirror those previously found to be sufficient to induce of long-lasting motor rescue with optogenetics. Experiments in Aim 1 will investigate the cellular mechanisms through which phasic stimulation in the STN evokes cell-type specific responses in the GPe (Aim 1.1) and use a machine learning approach to identify stimulation protocols that maximize this cell-type specific response (Aim 1.2). Experiments in Aim 2 will test the therapeutic efficacy of phasic stimulation protocols compared to conventional DBS, using behavioral and physiological assays in mouse (Aim 2.1) and primate (Aim 2.2) models of PD. Taken together, these experiments will advance our understanding of the fundamental differences between how conventional vs. phasic stimulation impacts the nervous system, with cell-type specific and synapse-specific resolution, and could provide novel therapeutic strategies that can be rapidly translated into humans.